天気予報はどれくらい当たるのか? 予報精度は年々向上している
Weather forecast or be hit much ? Forecast accuracy is improving every year

People midst of Golden Week , the weather forecast is a concern would be many . The weather forecast is , I wonder if is predicted with an accuracy of how on earth ?

■ hit rate of rain is " about 85% "
Tomorrow or become rain , hit or miss is the most worrisome in the weather forecast , it will either be sunny . There is a " chance of rain " in one of the forecast of rain . A chance of rain , I have shown " within a certain period of time , the probability of rain of 1 mm or more falls in the area ," the . Because it is the probability of rain or not rain a total of 1 mm or more or falls in that time period , it is not limited rainfall is more even 100 percent chance of rain . In addition, the chance of rain because it is published in 10% increments in practice (The 0% chance of rain means less than 5% chance of rain in some cases even chance of rain is 0% , 1 mm of rain less than that fall ) . No preference to be falling or continuous fall intermittently . In addition , everywhere is the same probability forecast area , it does not specify where in the area within or falls .

The chance of rain is calculated using a huge historical weather data Meteorological Agency has . For example , it is announced that when from the data of the past , to collect the cases similar to the state of the atmosphere , which is expected , " chance of rain 30 percent" , 30 times rain 100 times in that , was falling he has .

So, the hit or miss I wonder how much such ? The Japan Meteorological Agency , it is to verify the presence or absence of precipitation and precipitation probability of weather forecast , hit the edge of the maximum temperature , minimum temperature . In terms of the rate at which the " presence or absence of precipitation " was hit , at 85 % of the national average , has improved by about 5% in the last 20 years .

■ The forecast calculation , the effect of " Chaos " comes increased from about three days after
The weather forecast of modern , based on the laws of physics , was flipping super computer to the enormous computing " numerical weather prediction " is in the foundation . Accuracy of numerical weather prediction by the " global model " used for weather forecast has gone up every year , forecast error of the current "3 days in advance " is improved to the same extent as error of around 1985 "1 days in advance " have . However , long-term forecast ahead it is difficult . " Chaos " is because of the behavior appears in this atmosphere .

In numerical weather prediction , and there is a slight shift in the ( values ​​shown previously given to the computer when you start the forecast calculation , the state of the air current ) initial value , expected to go largely shifted while I repeated calculation . Of " chaos " this phenomenon . The chaos , the early 1960s , Edward Lorenz meteorologist ( 1917-2008 ) , was discovered when it had to unwind on the computer weather models . Chaos is a property that has air , it is not possible to resolve completely even to improve the technology of the forecast and observation .

Because of problems of chaos , in the long-term forecast , a method calculations are performed multiple times by changing slightly the initial value , and taking the average of them is taken ( ensemble forecast ) . The Japan Meteorological Agency (JMA) , you are doing and forecast for up to 1 week ahead , in the previous forecast of up to one month or more , the ensemble forecast .

■ I see that " the reliability of the weekly forecast" at the site of the Japan Meteorological Agency
In fact, it does not introduce to the weather forecast of television and newspapers , but in the web site of the Japan Meteorological Agency , for the presence or absence of rain for five days later in the week forecast the reliability of the predictions have been evaluated in three stages of A · B · C. The value of this confidence , I have used for the forecast of the 3rd beyond the uncertainty increases , was obtained by performing the ensemble forecast . In addition , instead of announcing been concluded to reflect the uncertainty of the forecast in the month forecast three month and forecast for one , " rain tomorrow ", " highest temperature is 25 ℃ " or , the results of the ensemble forecast based , and has published and so the temperature is of "consisting average year " is 30% , the " lower " is 20% , the " higher " and the 50% , .

Also , if you only change the numerical prediction of the atmosphere , the atmosphere by the " chaos " , the accuracy of the prediction falls . Sea 's a major impact on weather . Therefore , for the forecast to be more than one month the effects of ocean increases , also predicted (" [ phenomenon that sea surface temperature off Peru is increased every few years ] El Nino " to cause extreme weather events , for example) influence the sea on the air by the improvement in prediction accuracy is worn .

The weather forecast was announced for the first time in Japan it is June 1, 1884 . The effort to increase the accuracy current 130 years later , science and technology and meteorology has advanced by leaps and bounds he has been made.
天気予報はどれくらい当たるのか? 予報精度は年々向上している


■ 雨の的中率は「85%程度」


 では,その当たりはずれはどれくらいなのだろうか? 気象庁では,天気予報の降水確率や降水の有無,最高気温・最低気温の当たりはずれを検証している。「降水の有無」が的中した率でみると,全国平均85%程度で,この20年で5%程度向上している。

■ 予報計算は,およそ3日後から「カオス」の影響が大きくなってくる



■ 気象庁のサイトで「週間予報の信頼度」がわかる